Impact of foreign portfolio investments on market comovements: Evidence from the emerging Indian stock market Sunil Poshakwale and Chandra Thapa Cranfield School of Management, Cranfield University, Cranfield, Bedford, England, MK43 0AL Telephone: +44 (0) 1234 754404 Fax: +44 (0) 1234 752554 Email:[email protected]Paper to be presented in the Emerging Market Group ESRC Seminar on International Equity Markets Comovements and Contagion 11 May 2007 Cass Business School, London
Transcript
Impact of foreign portfolio investments on market comovements:
Evidence from the emerging Indian stock market
Sunil Poshakwale and Chandra Thapa Cranfield School of Management,
Paper to be presented in the Emerging Market Group ESRC Seminar on
International Equity Markets Comovements and Contagion 11 May 2007
Cass Business School, London
1
Impact of foreign portfolio investments on equity market comovements: Evidence from the emerging Indian stock market
________________________________________________________________________ Abstract This paper examines the influence of foreign institutional investments in explaining the short and
long run relationship of the Indian equity market with the main developed equity markets of the
US and the UK. Using daily return series and portfolio investments made by foreign institutional
investors and employing the VAR and Johansen’s cointegration framework, we find that the
mobility of foreign portfolio contains significant information in explaining the short and long
term comovements of the Indian equity market with that of the US and the UK equity markets.
We conclude that the rapid growth in the flow of the foreign portfolio investments is leading to
greater integration of the Indian equity market with the main developed markets and this may
have significant implications for asset pricing and international portfolio diversification benefits.
Impact of foreign portfolio investments on equity market comovement: Evidence from emerging Indian stock market
1. Introduction
Research documenting comovements amongst the international stock markets has been
growing. Many studies have been undertaken to identify important determinants of market
comovements (see for example, Hamao et al., 1990, Becker et al 1990, Lin, Engle and Ito, 1994,
Lognin and Solnik, 1995, Hsin, 2001, Bekaert and Harvey, 2003, W Dungey, 2004, Steeley,
2005, etc.). One of the main conclusions of the aforementioned studies is that the US and the UK
stock markets are often found to lead the market comovements because they transmit shocks not
only to their developed counterparts but also to the emerging stock markets in developing
countries. Dungey et al (2004) concludes that equity markets in Australia are affected by shocks
common to all other markets around the world. They find that the US market plays a significant
role in explaining the Australian equity market’s movement whilst Australia’s domestic output
has a very small impact that wanders away in the long run. Hsin (2004) examines the
comovements in stock indices amongst the major developed markets incorporating American,
European and Asia-Pacific developed markets. Consistent with previous findings, Hsin (2004)
concludes that US plays a significant part in transmitting shocks to the other markets. Hsin also
finds a strong linkage supported by evidence of transmission effects among the regional
participants in Europe, such as Germany, Britain and France and Asia-Pacific markets of Japan,
Australia, Hong Kong, and Singapore. On the other hand, Soydemir (2000) investigates the
pattern of comovements between developed and emerging market economies using the economic
fundamentals and trade linkages. He concludes that Mexico and USA show stronger linkages
whereas Argentina and Brazil reveal a weaker association and attributes this variation to the
trade flow differences. Soydemir also finds that the response to a shock from the US market has
3
longer effect than the one from the UK market. However, despite an overwhelming volume of
studies on transmission of information flow and volatility shocks, there is a lack of consensus on
whether it is economic fundamentals or contagion effects that cause comovements amongst the
international stock markets. For instance, King and Wadhwani (1990) argue that in tracking
market movements, most international investors ignore the fundamental factors and only track
the mobility of some leading markets such as the US stock market.
It is widely known that emerging economies have been undertaking reform activities
primarily to attract foreign capital to meet the growing demand for investment in infrastructure,
institution building, etc. In his paper, Errunza (2001) suggests that the increase in foreign
investment portfolio augments market integration further supplementing information to explain
equity market comovements. Despite the rapid economic growth witnessed in the emerging
markets, there is very little research that has been devoted in understanding the role and impact
of the foreign portfolio investments on the comovements of the emerging equity markets with the
developed stock markets. The research on role of portfolio investment flows is of topical interest
particularly in light of the empirical evidence that portfolio investment flows to developing
countries have been on the rise and exceeded US$100 billion in 2005 (UNCTAD, 2006). As
suggested by Soydemir (2000), if flow of foreign capital integrates the transitional economies
with the world, the capital mobility itself should be one of the fundamental sources influencing
their comovements. This paper aims to address this gap by documenting evidence from the
emerging Indian stock market by investigating the relationship between foreign portfolio
investment flows and its impact on the short and long run comovements with the two major
developed stock markets i.e., US and the UK.
4
Research on emerging stock markets like India has important theoretical and policy
implications. Emerging markets such as India have long been viewed by international investors
as segmented markets offering excellent diversification benefits to international investors
because of their relatively low correlation with developed stock markets. However, if it is found
that the Indian stock market is well integrated with the main developed markets, then this would
imply reduced diversification benefits. In this context previous empirical research by Bekaert
and Harvey (2000) has made attempts to investigate the implications of integration of emerging
stock markets with the global markets using asset pricing framework. Other researchers such as
Lane and Milesi-Ferretti (2003) have done research highlighting the impact of foreign portfolio
and direct investments on the financial integration for a sample of industrialized countries. There
are relatively fewer studies involving emerging markets and even fewer on the emerging Indian
stock market. For example, Lamba (2005) using data from July 1997 to December 2003 reports
that the Indian market is being increasingly influenced by the US and UK market and their
impact have been persistent since the September 11 attack in the US. Sharma (2003) investigates
the impact of foreign investment on India’s export performance and finds no statistically
significant relationship between the two. However, research that directly relates the role of
foreign institutional investor’s buying activities on comovements of the Indian emerging equity
market with its developed counterparts is lacking.
This paper examines the influence of foreign institutional portfolio investments in
explaining the short and long run relationship of the Indian equity market with the main
developed equity markets of the US and the UK. Using daily return series and portfolio
investments made by foreign institutional investors and employing the VAR and Cointegration
tests, we find that the mobility of foreign portfolio contains significant information in explaining
5
the short and long term comovements of the Indian equity market with that of US and the UK
equity markets. We conclude that the rapid growth in the flow of the foreign portfolio
investments is leading to greater integration of the Indian equity market with the main developed
markets and this may have significant implications for asset pricing and international portfolio
diversification benefits.
The paper is organized as follows. The following section explains the data and
methodology. Section 3 reports the empirical findings, and section 4 concludes the paper.
2. Data and Methodology 2.1 Data
We use daily data in our analysis for a sample period of six years beginning 1 January
2001 through 15 January 2007. We calculate equity returns from the MSCI price index for US,
U.K and India. The MSCI index time series have been obtained from Datastream international.
Net daily investment data of the portfolio investments made by foreign institutional investors is
obtained from CNBC’s Moneycontrol.com.
2.2 Methodology
2.2.1 Vector Autoregression
We employ a Vector Autoregressive model to investigate the impact of foreign portfolio
investor on comovement of the Indian economy. A non-structural modeling approach developed
by Sims (1980) such as VAR is widely used in investigating relationship among economic
variables especially because structural frameworks require economic theory which is often not
rich enough to provide a dynamic specification adequate to identify relationships among
6
variables. The VAR is an effective framework for forecasting systems of interrelated time series
and enables analysis of the dynamic impact of shocks on the system of variables. The VAR
approach also overcomes the need for structural modeling by treating every endogenous variable
in the system as a function of the lagged values of all endogenous variables considered in the
system and is specified as follows:
ttptptt BxyAyAy ε++++= −− .........11 (1)
where yt is a K vector of endogenous variables, xt is a d vector of exogenous variables, A1,
……….Ap and B are matrices of coefficients to be estimated, and itε is a vector innovations that
may be contemporaneously correlated with all of the right-hand side variables.
We estimate the following OLS model in VAR system:
tRi, = iα + ∑=
−I
iLtiiR
1,β + ∑
=
I
jtjjR
1,γ + ti
I
iiFFC ,
1∑=θ + itε (2)
Where,
tRi, is return on country i (i.e., India) at time t.
Ri,t-L is return on country i for lag t-L
Rj,t is return on country j at time (i.e., US and UK).
FFCi,t is flow of foreign capital for country i at time t.
itε is the white noise error term. In our model, since only lagged values of the endogenous variables appear on the right
hand side of the equations, simultaneity is a non-issue. The Ordinary Least Squares (OLS) in
VAR framework provide consistent and efficient estimates even though the error term ( itε ) may
be contemporaneously correlated because all equations have identical regressors.
7
2.2.2 Impulse Response Function The F statistic provided by VAR by construction does not facilitate identification of the
sign of relationship (i.e. whether changes in one return index will have negative or positive effect
on the other return variable). Also, VAR is not able to explain how long it takes for the change in
one variable permeating through the system to have an effect on the other variable. Impulse
response function identifies the responsiveness of one factor in the VAR system to shocks or
innovations of each of the other variables. It explains how a unit shock in one variable in
isolation of the others, affects the movement in other variables. In each of the equation one unit
shock is applied to detect the change in the VAR system over time by representing the VAR as
VMA (Vector Moving Average) representation:
tRi, = 0
11b ti ,ε + 111b 1,1 −tε + 1
12b 1,2 −tε +…….. (3)
Where, bij are unit normalized innovation coefficients of impulse response function
following the normalization by the Cholesky factor1 and 011b is the simultaneous effect of a unit
shock to ti ,ε . The contemporaneous innovation is stated in standard deviation form and have non-
unit coefficient in contrast to its unit coefficient in equation
2.2.3 Variance Decomposition Analysis Previous research has shown that variance decomposition analysis is an effective way to examine
the dynamic interactions amongst economic time series. Whilst impulse response function traces
the effects of a shock to one endogenous variable on to other variables in the VAR, variance
decomposition enables further analysis by separating the variation in an endogenous variable.
1 See Diebold (2004).
8
The variance decomposition thus offers greater insights about the relative significance of each
random innovation that affects the variables in VAR. Decomposing the variance offers slightly
different perspective on the relationship of the identified variables since it shows what proportion
of the variance is due shock of its own lags against the shocks of other variables. A shock to
variable i will not only affect its own future outcomes but will also be transmitted to other
variables. Variance decomposition provides the magnitude of the effect on itself and others in the
system. Studies have shown that most of the shocks in the h-step error variance are attributed to
its own shocks.
2.2.4 Cointegration
We examine the long run relationship between Indian, US, UK equity markets and net
flow of foreign institutional investment using the VAR analysis proposed by Johansen (1988)
and Johansen and Juselius (1990). We follow Johansen-Juselius (JJ) because their approach is
considered superior to the regression-based approach suggested by Engle and Granger in 1987
(Cheung and Lai, 1993).2 Another reason for using the JJ approach is that it utilizes the
maximum likelihood estimates and allows testing and estimation of more than one cointegrating
vector in the multivariate system without requiring a specific variable to be normalised. This
way, the JJ tests overcome the problem of carrying over the errors from the first step into the
second step commonly encountered in Engle and Granger’s (1987) approach to cointegration.
Further, Johansen’s method is independent of the choice of the endogenous variable within a
vector autoregression (VAR) framework which enables testing for various structural hypotheses
2 The Johansen-Juselius procedure resolves the problem of endogeneity in that we do not need to normalise the cointegrating vector on one of the variables as required in the Engle and Granger (EG) test.
9
involving restricted versions of cointegrating vectors and speed of adjustment parameters using
likelihood ratio tests.
The VAR equation in 2.2.1 can be rewritten as,
∑−
=−− ++∆Γ+Π=∆
1
11
p
ittititt Bxyyy ε (4)
Where:
∑=
−=Πp
ii IA
1, and (5)
∑+=
−=Γp
ijji A
1
(6)
Since our objective is to investigate the long-run relationship, we will focus on the
elements of matrix Π. If vector y contains m variables, matrix Π will be of the order m x m, with
a maximum possible rank of m (or full rank). Equation (4), except for the Πyt-k term, is in the
form of the traditional VAR with first difference. The Π term determines whether the system of
equations is cointegrated, i.e., whether a long-run equilibrium relationship exists. The feature to
note is that the rank of matrix Π is equal to the number of independent cointegrating vectors. If
rank of matrix Π = 0, the matrix is null, i.e., all the elements in this matrix are zero, which
implies no cointegration or in other words lack of a long-run equilibrium relationship and the
error correction mechanism, Πyt-k, therefore, does not exist. In determining the rank of matrix Π
(number of cointegrating vectors), we calculate the characteristic roots or eigenvalues, iλ̂ of Π.
Johansen (1988) and Johansen and Juselius (1990) propose trace (λtrace) and maximum
eigenvalue (λmax) test statistics to establish whether the characteristics roots are significantly
different from zero. The likelihood ratio statistic for the trace test (λtrace) is:
10
( )∑+=
−−=m
riitrace Tr
1
ˆ1ln)( λλ (7)
where iλ̂ = the estimated values of the characteristic roots (also known as
eigenvalues) obtained from the estimated Π matrix. The null hypothesis to be tested is that the
number of cointegrating vectors is less than or equal to r against the alternative hypothesis that
the number of cointegrating vectors is more than r. For example the null hypothesis r ≤ 0 against
alternative r = 1, r ≤ 1 against alternative r = 2, and so forth. The computed values of λtrace
statistics are evaluated using the critical values provided by Osterwarld-Lenum (1992). The
optimal system lag length is determined by using the Schwarz Information Criteria (SIC).
Specifically, the appropriate number of lags for each variable is obtained by computing the SIC
over different lag schemes in the range from 1 to 20 and by choosing the number of lags that
yields the lowest value for the SIC. We report the eigenvalue and trace statistic, with the
interpretation of the latter to investigate the long term relationship.
2.2.5 Granger Causality
The short-term relationship is examined using the Granger causality between the
endogenous variables in the following way:
Ry,t = a + ∑=
n
i 1βi Ry,t-i + ∑
=
n
i 1γi Rx,t-i + ε t (8)
Rx,t = a + ∑=
n
i 1δi R x,t-i + ∑
=
n
i 1ζi Ry,t-ι + u t (9)
where Ry,t and Rx,t are the returns of index y and x at time t accordingly.
In the above regressions we examine whether the coefficients γi and ζi are equal to zero
using a standard F test. If γi, and ζi coefficients are different from zero then we conclude that there
11
is a bi-directional causality between and Ry,t and Rx,t. Alternatively, if both coefficients are
found to be equal to zero, then we are able to conclude that there is no causality. Finally, in
equation (9) Ry,t Granger causes Rx,t if γi =0 for i=1,2,…n. Similarly, in (10) causality implies
that Rx,t Granger causes Ry,t , provided that ζi ≠ 0 for i=1,2,…n. Our causality test also uses FII
data to test the causality with returns.
3. Empirical Results
Figure 1 presents the time series plot of log values of price index (price) and log of net
investments made by foreign institutional investors (FII). In isolation of other confounding
factors, intuitively the figure indicates that there is a strong association between the flow of
foreign portfolio investments and the movement of the Indian stock price index (correlation
coefficient of 0.9). Figure 2 presents the plot of gross monthly portfolio investments by US
investors which on average constitute 16% of the total gross portfolio investments during the
period of January 2001 to November 20063.
In table 1, we provide descriptive statistics for all time series considered in our analysis.
Data presented in Table 1 shows that average daily returns from the Indian market are higher
than those offered by the US or the UK. However as expected, higher returns come with
relatively higher risk in the Indian market as confirmed by higher standard deviation of 1.46 %
as opposed to 1.12% for the US and 1.08% for the UK. Further, we find that kurtosis for all the
return series and foreign flow of investment are above 3 which confirm that distribution of return
is not normal as the Jarque-Bera statistic is significant in all cases.4 This is further confirmed by
3 See CNBC, MoneyControl.com and U.S. Treasury International Capital 4 One of the assumptions of the VAR which uses the OLS estimation method is violated. However, since sample sizes are sufficiently large, the violation from the normality assumption does not significantly affect the estimation. Following the central limit theorem, the statistics will asymptotically converge to normality.
12
skewness statistic which shows that the US and Indian returns are negatively skewed. However,
daily return from the UK are positively skewed suggesting that on average and in terms of
frequency greater positive returns were available from the UK market during the sample period.
It is a common practice to investigate whether the time series are stationery in its level
form. For testing this, we use the Augmented Dickey Fuller (ADF) test. The results reported in
Table 2 show that all series in the logarithmic level form are non-stationery as confirmed by the
insignificant ADF statistic. However, the first differenced series are integrated to the order I (1)
with ADF statistic being significant at 5% level. Hence, all series are transformed to log of first
difference for the VAR analysis and level data are used for the cointegration analysis.
Short Run Relationship
As explained in section 2.2.5, the Granger causality model can be effectively used in
investigating for the short term causal interrelationships between variables. It gives us a practical
approach to assess whether short term movements in daily equity returns and foreign investments
are related. It also indicates the direction of causality, i.e. whether there is unidirectional, or
bidirectional, or no causality.
Results reported in Table 3 demonstrate that the null hypothesis that changes in the
portfolio investment flows do not cause changes in the Indian stock price index returns and vice
versa can be rejected at 5 per cent level of significance. The results also show that the UK and
the US markets seems to Granger cause the portfolio investment flows into India. Consistent
with previous findings, the US and the UK markets do seem to influence the movements in the
Indian stock market in the short run and also influence the flow of foreign capital. The findings
13
lend support to the hypothesis that that the foreign portfolio investments are contributing to the
greater integration of the Indian equity markets with equity markets of the US and the UK.
Long Run Relationship
Table 4 presents the eigenvalues and the trace statistics estimated by employing Johansen
and Juselius maximum likelihood estimation. The cointegration tests examine that there are r
cointegrating relationship against the r + 1 and hence, not accepting the null implies that the
cointegrating relationship exceeds r number of relationships by one. Results in Panel A show
that most trace statistics are significant at 1% level leading us to reject the null hypothesis that
there is no cointegrating relationship between daily equity returns from India, US, and UK as
well as the portfolio investment flows. Since we are interested in investigating the relationship of
foreign portfolio investment flows and the Indian stock market, we employ a bivariate
cointegration test which shows that the null hypothesis of no cointegration between the daily
movements in the Indian equity markets and portfolio investment flows can be rejected at 1%
and 5%. Thus, results so far seem to confirm that foreign portfolio investment flows not only
influence daily movements in the Indian equity market in the short run but they also seem to
influence the long-run returns.
Impulse Response Function
Figure 3-8 show the impulse responses and their upper and lower level bands. The
impulse response is considered statistically insignificant if the upper and lower level bands cross
the horizontal axis. Figure 3 represent the response of innovations in foreign portfolio investment
flows (FII) to its own innovations. It is clear that the peak effect occurs on day one and remains
significant for up to a week. However, neither the upper nor the lower band crosses the
14
horizontal axis indicating that changes in the portfolio investment flows are significantly
influenced by shocks in its own time series.
Figure 4 show how innovations in the FII influence the Indian stock market returns. The
upper and lower bands do not cross the horizontal axis and thus confirm that innovations in the
FII significantly influence the Indian stock market returns. The response of the Indian market to
the foreign portfolio investment flows is highest on day one which gradually declines until the 6th
day before it fades off.
Figure 5 presents the response of the FII to the innovations in the Indian stock market
returns. The impulse response function is once again significant peaking on day 2 before it starts
to decline and fades off from day 6. This suggests a positive feedback trading by international
investors and is consistent with those reported by previous studies (see, Froot et al., 2001).
Figure 6 shows the response of the Indian stock market returns to its own shocks. The
peak effect occurs on day one which remains significant on day 2. The lower bound crosses the
horizontal axis between day 2 and 3 and the upper bound follows on day 3 suggesting that the
response of the Indian stock returns to its own innovations is short lived and only lasts for couple
of days. The dispersion around the mean response is also quite small.
Figure 7 shows the response of the Indian stock market to the changes in the UK stock
market. The Indian stock market response seems to peak on day 2 before it begin to decline
suggesting that the market responds to the news and events in the UK market one day later which
is consistent with the time differences in trading hours between Indian and the UK stock markets.
Figure 8 presents the innovations in the US markets and the response of the Indian stock
market. The upper and lower responses are closer to the mean response and also the lower bound
crosses the horizontal axis on day 2. This suggests that the response coming in from the US
15
market is somewhat less significant than those observed in the case of UK stock market shocks
to the Indian stock market. However, consistent with the results in figure 7, the peak response of
the Indian market to the innovations in the US market occurs on day 2 (one day late) which
declines and fades away rapidly.
Overall, the impulse response tests confirm that movements in the US and the UK
markets do seem to impact the movements in Indian stock market and more significantly they are
consistent with those observed for the innovations in net portfolio investment flows and response
of the Indian market to those innovations. The short term impact in all figures appears
significant. Our results are robust because the confidence bands around the impulse response
functions are calculated by using the Monte Carlo method. The results are consistent with the
findings of the Granger causality test and lead us to conclude that the US and the UK markets not
only influence the movement of Indian stock market but they also significantly affect the
mobility of portfolio investment flows in India. This suggests that the comovements in the Indian
equity market are explained by the net portfolio investment flows by foreign institutional
investors who seem to be influencing the integration process of the Indian stock markets with the
developed equity markets of the US and the UK.
Variance Decomposition
A further tool is available in the VAR system which uses the recursive scheme so that the
MA representation mentioned in the above section with white noise innovations could also be
utilized to examine the degree of variance in the h-step error due to its own innovation and
innovation in other variables.
16
Table 5 and 6 show the forecast error variance decompositions of return from Indian
market and portfolio investment flows. Our results do not seem to be affected by changing the
order factorization, despite the possibility of the error variance of the VAR system being
contemporaneously correlated, particularly due to the high correlation between US and UK
returns. However, since the main purpose of our study is to examine the influence of the foreign
investment portfolio flows and innovations in the UK and the US stock markets on the Indian
stock market, the contemporaneous relationship of the VAR residuals owing to the high
correlation between US and UK returns will not significantly influence the interpretation of the
results. Results in Table 5 show that the fraction of error variance in forecasting returns in the
Indian stock market due to innovation in the foreign institutional portfolio investments is zero in
the one step horizon but gradually builds in long horizon which reflects that it takes time to
transmit the effects of information contained in the foreign institutional portfolio investment
flows into the Indian stock market returns. Since we use daily data, even 0.7% in the 5 step
horizon could be considered quite significant. We also observe that both the US and the UK
market carry relatively substantial information in sharing the h-step ahead variance in the Indian
stock market. The US return accounts for almost 3% error variance right from the second period
and it gradually increases with time.
In table 6, we observe that FII future error variance is accounted for by the innovations in
return in Indian, the US and the UK stock markets which suggest that foreign institutional
investors are return chasers. The results also confirm that the movements in returns from the US
and the UK contribute to the error variance in returns from the Indian stock market. Thus we
conclude that Indian returns are not only influenced by returns in the US and the UK but the
foreign institutional portfolio investment flows play a significant part in explaining the
17
comovements of the Indian equity market with the developed equity markets of the US and the
UK.
Additional Tests
In order to further support our above result, we ran two separate bivariate regressions
using the logarithmic level data. We regress the Indian log index values on the index values of
the US and the UK. The residuals from the two bivariate regressions, which are non stationary
(as confirmed by the ADF test) and linear combination of the two markets (proxy for comovment
and long run relationship) are then used as exogenous variables to investigate how well they
explain the foreign investment portfolio flows (FII). Results (not reported here but available on
request) are highly significant, which further complement the empirical results obtained through
VAR, Impulse Response Function and the Variance Decomposition analysis and confirm that the
foreign portfolio investment flow significantly influence the comovements of the Indian market
with that of US and the UK.
5. Conclusions
This study examines the comovements of the Indian stock market with the major developed
stock markets of the US and the UK and the how the foreign portfolio investment flows in the
Indian stock market influence this process. Using VAR (Impulse Response Function and
Variance Decomposition) analysis, on daily data from January 2001 till 15 January 2007, we find
that Indian stock returns are significantly influenced by the short and long term innovations in
the US and UK stock markets. The short term and long term relationship of the Indian stock
market with the developed UK and the US stock markets as well as foreign portfolio investment
flows is confirmed using the Granger causality and Johansen and Jesulius Cointegration tests.
18
Further examination of the robustness of the results is achieved by employing the impulse
response analysis which confirms that the Indian stock returns respond significantly to the
innovations in the UK and the US stock markets as well as to the innovations in foreign portfolio
investment flows. Further, variance decomposition analysis results suggest that both the US and
the UK markets carry relatively substantial information in sharing the h-step ahead variance of
the Indian Stock Market and FII future error variance is accounted for by the innovations in
equity returns in the Indian, the US and the UK stock markets. Our results thus suggest that the
impact of the foreign portfolio investment flows on Indian stock returns is significant and these
flows seem to explain and influence the short and long run comovements of the Indian stock
markets with the developed stock markets of the US and the UK.
19
References
Becker, K. G., Finnerty, J. E. and Gupta, M. (1990), "The Intertemporal Relation Between the U.S. and Japanese Stock Markets", The Journal of Finance, vol. 45, no. 4, pp. 1297.
Bekaert, G. and Harvey, C. R. (2003), "Emerging markets finance", Journal of Empirical Finance, vol. 10, no. 1/2, pp. 3.
Campbell, John Y., Andrew W. Lo and A. Craig MacKinlay (1997), The Econometrics of Financial Markets, Princeton University Press.
Chelley-Steeley, P. L. (2005), "Modeling equity market integration using smooth transition analysis: A study of Eastern European stock markets", Journal of International Money and Finance, vol. 24, no. 5, pp. 818.
Cheung, Y. and Lai, K. S. (1993), "A fractional cointegration analysis of purchasing power
parity", Journal of Business & Economic Statistics, vol. 11, no. 1, pp. 103.
Chowdhury, A. R. (1994), "Stock market interdependencies: Evidence from the Asian NIEs", Journal of Macroeconomics, vol. 16, no. 4, pp. 629.
Dungey, M., Fry, R. and Martin, V. L. (2004), "Identification of common and idiosyncratic shocks in real equity prices: Australia, 1982-2002", Global Finance Journal, vol. 15, no. 1, pp. 81.
Errunza, V. (2001), "Foreign portfolio equity investments, financial liberalization, and economic development", Review of International Economics, vol. 9, no. 4, pp. 703.
Eun, C. S. and Shim, S. (1989), "International Transmission of Stock Market Movements", Journal of Financial and Quantitative Analysis, vol. 24, no. 2, pp. 241.
Francis X Diebold (2004), Elements of Forecasting, 3rd ed, South Western.
Froot, K. A, O’connel, P.G.J and Seasholes, M.S (2001), “The portfolio flows of international investor”, Journal of Financial Economics, vol. 59, pp. 151.
Hamao, Y., Masulis, R. W. and Ng, V. (1990), "Correlations in Price Changes and Volatility across International Stock Markets", The Review of Financial Studies (1986-1998), vol. 3, no. 2, pp. 281.
Hsin, C. (2004), "A multilateral approach to examining the comovements among major world equity markets", International Review of Financial Analysis, vol. 13, no. 4, pp. 433.
Johansen, S. and Juselius, K. (1990), "Maximum Likelihood Estimation and Inference on
Cointegration - With Applications to the Demand for Money", Oxford Bulletin of Economics and Statistics, vol. 52, no. 2, pp. 169.
20
King, M. A. and Wadhwani, S. (1990), "Transmission of Volatility between Stock Markets", The Review of Financial Studies (1986-1998), vol. 3, no. 1, pp. 5.
Lamba, A. S. (2005), "An Analysis of the Short- and Long-Run Relationships Between South Asian and Developed Equity Markets", International Journal of Business, vol. 10, no. 4, pp. 383.
Lamba, A. S. (1999), "How isolated is the Indian stock market empirical research findings?", Journal of Financial Management & Analysis, vol. 12, no. 1, pp. 35.
Lane, P. R. and Milesi-Ferretti, G. M. (2003), "International financial integration", IMF Staff
Papers, vol. 50, pp. 82.
Lin, W., Engle, R. F. and Ito, T. (1994), "Do Bulls and Bears Move across Borders? International Transmission of Stock Returns and Volatility", The Review of Financial Studies (1986-1998), vol. 7, no. 3, pp. 507.
Longin, F. and Solnik, B. (2001), "Extreme correlation of international equity markets", The Journal of Finance, vol. 56, no. 2, pp. 649.
Sims, C. A.(1980), “Macroeconomics and reality”, Econometrica, 48, pp. 1-49. Soren, J (1991), "Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models", Econometrica (1986-1998), vol. 59, no. 6, pp. 1551.
Soydemir, G. (2000), "International transmission mechanism of stock market movements: Evidence from emerging equity markets", Journal of Forecasting, vol. 19, no. 3, pp. 149.
Svetlozar T. R., Stefan M, Frank J. F., Serg (2007), “Financial Econometrics: From Basics to Advance Modeling Technique” John Wiley and Sons, Inc.
21
Table 1 Descriptive
IND UK USA FII Mean 0.0839 0.0033 0.0213 1062.993 Median 0.1608 0.0053 0.0320 564.500 Maximum 8.6149 5.7713 5.3966 34905.000 Minimum -11.2642 -5.0738 -5.1239 -3075.000 Standard Deviation 1.4604 1.0786 1.1256 3425.665 Skewness -0.5953 0.1859 -0.1869 0.4380 Kurtosis 8.6150 5.9760 5.6077 30.1947 Jarque-Bera 1908.123 520.946 401.9250 42876.740 Probability 0.0000 0.0000 0.0000 0.0000
Note FII is the daily net purchase by foreign institutional investor (in millions local currency) IND, USA and UK are daily returns (in %) from India, US, and the UK MSCI indices.
Table 2 ADF Test Statistic
Level Data First Difference
With intercept and no trend
With intercept and trend
With intercept and no trend
With intercept and trend
Critical Value (1%) -3.434275 -3.963851 -3.434278 -3.963855 Critical Value (5%) -2.863161 -3.412651 -2.863162 -3.412653 USA 1.203393 -1.999717 -40.9302 -41.02679 UK -0.451379 -2.150349 -40.9334 -40.98671 IND 2.415804 -0.843257 -28.0793 -28.21655 FII 3.230412 -0.819944 -25.9890 -26.15281
22
Table 3 Pair wise Granger Causality Tests
Null Hypothesis F-Statistic Probability IND does not Granger Cause FII 10.1076 0.0000 FII does not Granger Cause IND 4.03195 0.01792 IND does not Granger Cause USA 0.56311 0.56955 USA does not Granger Cause IND 9.11082 0.00012 IND does not Granger Cause UK 0.88006 0.41496 UK does not Granger Cause IND 28.9015 0.0000 FII does not Granger Cause USA 2.44604 0.08696 USA does not Granger Cause FII 5.2678 0.00525 FII does not Granger Cause UK 1.40781 0.24499 UK does not Granger Cause FII 15.9834 0.0000
Table 4
Panel A: Cointegration Test (U.S.A, U.K., FII and IND) Critical Value Null Hypothesis Eigenvalue Trace Statistic 5% 1% r ≤ 0 0.019788 54.65325 47.21 54.46 r ≤ 1 0.014943 30.05022 29.68 35.65 r ≤ 2 0.005983 11.51601 15.41 20.04 r ≤ 3 0.003348 4.128526 3.76 6.65
Panel B: For series FII and IND only
Critical Value Null Hypothesis Eigenvalue Trace Statistic 5% 1% r ≤ 0 0.013170 22.43145 15.41 20.04 r ≤ 1 0.004952 6.111430 3.76 6.65
Table 5 India First Ordering Second Ordering Period FII IND UK USA FII IND UK USA 1 0 96.69 0.02 3.29 0 96.69 0.83 2.48 2 0.32 92.92 2.16 4.61 0.32 92.92 4.18 2.59 3 0.63 92.40 2.40 4.56 0.63 92.4 4.41 2.56 4 0.65 92.36 2.41 4.59 0.65 92.36 4.43 2.57 5 0.65 92.35 2.41 4.59 0.65 92.35 4.43 2.57
23
Table 6 FII First Ordering Second Ordering Period FII IND UK USA FII IND UK USA 1 96.43 3.32 0.07 0.18 96.43 3.32 0.17 0.08 2 93.06 4.38 1.45 1.12 93.06 4.38 2.35 0.21 3 92.06 4.87 1.75 1.31 92.06 4.87 2.83 0.24 4 91.79 4.97 1.88 1.36 91.79 4.97 3.00 0.24 5 91.72 5.00 1.90 1.38 91.72 5.00 3.03 0.24
Ordering 1: USA,UK, IND, FII Ordering2: UK, USA, IND, FII
Fig 1
Net foreign institutional investors' purchase and Index
0.00
2.004.00
6.00
8.00
10.0012.00
14.00
1/1/
2001
7/1/
2001
1/1/
2002
7/1/
2002
1/1/
2003
7/1/
2003
1/1/
2004
7/1/
2004
1/1/
2005
Daily Observation (01/01/2001 - 15/01/2007)
loga
rithm
ic s
cale
IndexLog-Nfii
Fig 2
U.S vs Total Purchase
-
50,000.00
100,000.00
150,000.00
200,000.00
250,000.00
2001
-01
2001
-09
2002
-05
2003
-01
2003
-09
2004
-05
2005
-01
2005
-09
2006
-05
Year-Month
USD
in M
illio
ns
Purchase by U.S FIITotal purchase by FII
24
Average Monthly Gross Purchase from January 2001 to November 2006
Year-Month
U.S. Purchase ( Millions USD)
Total Gross Purchase(Million USD) U.S share in %
Total 525 3,230 16.27
Source: CNBC, MoneyControl.com and U.S. Treasury International Capital
